Abstract: A method in a computing environment for application start up includes: generating a directed acyclic graph (DAG) for tasks that are performed during application start-up; performing topological sorting of the tasks based on the DAG to define an order for the tasks; allocating a tenant-specific priority queue; loading the tasks in the tenant-specific priority queue in the order determined by the topological sorting; allocating a plurality of computing threads to executing the tasks; executing the tasks using the plurality of computing threads that were allocated, wherein a plurality of the tasks are executed in parallel, and the tasks are assigned to the computing threads in an order based on the order in which the task are loaded in the tenant-specific priority queue; preparing a start page for use in accessing the application; and causing the start page to be displayed on a user access device for accessing the application.

Abstract: In some embodiments, a method determines a first functional domain that includes a group of security policies that have been copied from a second functional domain. Network flow data is queried to determine network traffic that is associated with a security policy in the group of security policies in the first functional domain. The method analyzes utilization of the security policy based on the network traffic. Based on the analyzing, a recommendation is generated to change the security policy in the first functional domain.

Abstract: In various embodiments, a data storage system maintains a data store for a plurality of groups of users, where the data store includes a first section that is accessible by a first group of users, but not a second group of users, via the data storage system. In response to receiving, from a first user of the first group of users, a request to share data with the second group of users, the data storage system sends a request to a second user of the second group of users. Subsequent to receiving an acceptance from the second user, the data is stored in a second object in a second section of the data store that is accessible by the second group of users, but not the first group of users, via the data storage system.

Abstract: Techniques for modifying a period of time that data, associated with a characteristic, transmitted via a communication platform is retained are described. A data retention rule can include a first period of time for retaining data transmitted via the communication platform, in association with an organization. The first user can additionally establish a specific data retention rule associated with data associated with a particular characteristic. The specific data retention rule can include an instruction to store communications including the particular characteristic for a second time period that is different from the first time period associated with the data retention rule. The communication platform can receive data from a second user computing device associated with the second user of the organization. Based on a determination that the data is associated with the characteristic, the communication platform can store the data according to the data retention rule specified for such data.

Abstract: Apparatus and method for identifying code path contributors to resource utilization spikes. For example, in some implementations, stack trace data and profiling data are sampled from a plurality of code paths of an application to generate a plurality of code path samples, where the sampling is performed across a series of temporal windows. Processor utilization data is also sampled across a corresponding series of temporal windows. Correlations are detected between the code path samples and occurrences of processor utilization spikes to generate correlational scores for the plurality of code paths. A ranking of code paths is then generated in accordance with the correlational scores in which code paths relatively higher in the ranking are more likely to contribute to the occurrences of the processor utilization spikes than code paths relatively lower in the ranking.

Abstract: A method and system for rendering a stack trace visualization display has been developed. A first stack trace associated with execution of an application during a time period is received from a central processing unit profiler. A first stack trace visualization display is rendered including a plurality of stack frames stacked in accordance with an order of ancestry based on the first stack trace. Rendering at least one stack frame involves rendering at a first location of the first stack trace visualization display, a stack frame rectangle for the at least one stack frame in accordance with the order of ancestry and rendering at a second location of the first stack trace visualization display, stack frame specific text for the at least one stack frame. The second location overlays the first location. Rendering of the stack frame rectangle is independent of the rendering of the stack frame specific text.

Abstract: A field extraction system that does not require field-level annotations for training is provided. Specifically, the training process is bootstrapped by mining pseudo-labels from unlabeled forms using simple rules. Then, a transformer-based structure is used to model interactions between text tokens in the input form and predict a field tag for each token accordingly. The pseudo-labels are used to supervise the transformer training. As the pseudo-labels are noisy, a refinement module that contains a sequence of branches is used to refine the pseudo-labels. Each of the refinement branches conducts field tagging and generates refined labels. At each stage, a branch is optimized by the labels ensembled from all previous branches to reduce label noise.

Abstract: A client application and a local security controller (LSC) executing on a host computing device use a Multiparty Computation (MPC) cryptographic key generation technique to create two fragments of a split private key, which are held by the client application and LSC, respectively. The client application generates a certificate signing request (CSR). The client application and LSC sign the CSR with the split private key using an MPC technique. The LSC then signs a token from the client application to indicate that the private key corresponding to the CSR is MPC-backed. A package with the CSR and the first and second signatures is then sent to a remote device acting as a certificate authority. The remote device verifies the two signatures and issues a certificate to the client application. The second signature is verified using information sent to the remote device from the LSC during a registration process.

Abstract: A method is provided for generating regular expressions. In some embodiments, the method includes partitioning a first plurality of text strings into a plurality of substrings, each text string of the first plurality being malignant or benign. The method also includes generating a population list including a second plurality of text strings, each text string of the second plurality including at least one token. The method further includes creating subsequent iterations of the population list. Each iteration may include calculating a score for one or more text strings in the population list and removing one or more text strings from the population list. The method further includes in response to a determination to not update the population list, selecting, based on the one or more calculated scores, a third text string as a regular expression from a final iteration of the population list.

Abstract: A method for using a neural network model for natural language processing (NLP) includes receiving training data associated with a source domain and a target domain; and generating one or more query batches. Each query batch includes one or more source tasks associated with the source domain and one or more target tasks associated with the target domain. For each query batch, class representations are generated for each class in the source domain and the target domain. A query batch loss for the query batch is generated based on the corresponding class representations. An optimization is performed on the neural network model by adjusting its network parameters based on the query batch loss. The optimized neural network model is used to perform one or more new NLP tasks.

Abstract: Techniques are disclosed relating to automated operations management. In various embodiments, a computer system accesses operational information that defines commands for an operational scenario and accesses blueprints that describe operational entities in a target computer environment related to the operational scenario. The computer system implements the operational scenario for the target computer environment. The implementing may include executing a hierarchy of controller modules that include an orchestrator controller module at top level of the hierarchy that is executable to carry out the commands by issuing instructions to controller modules at a next level. The controller modules may be executable to manage the operational entities according to the blueprints to complete the operational scenario.

Abstract: A database with virtual partitioning. A computer system used to implement the database receives a request for a database operation to be performed on data stored in a portion of the database, where the data has a group of logical partitions (e.g., tenants of a multi-tenant database), and where the portion of the database is not further physically partitioned. The computer system identifies a virtual partitioning scheme for the database operation, the virtual partitioning scheme defining sub-groups of the group of logical partitions. The computer system then performs the database operation for the sub-groups defined by the virtual partitioning scheme. Multiple virtual partitioning schemes can be active at the same time, and virtual partitioning schemes can be created dynamically in some instances.

Abstract: Embodiments described herein propose an approach for unsupervised structure extraction in task-oriented dialogues. Specifically, a Slot Boundary Detection (SBD) module is adopted, for which utterances from training domains are tagged with the conventional BIO schema but without the slot names. A transformer-based classifier is trained to detect the boundary of potential slot tokens in the test domain. Next, while the state number is usually unknown, it is more reasonable to assume the slot number is given when analyzing a dialogue system. The detected tokens are clustered into the number of slot of groups. Finally, the dialogue state is represented with a vector recording the modification times of every slot. The slot values are then tracked through each dialogue session in the corpus and label utterances with their dialogue states accordingly. The semantic structure is portrayed by computing the transition frequencies among the unique states.

Abstract: Artificial intelligence (AI) technology can be used process natural language statements to facilitate the automated generation of narratives about data sets that achieve a desired communication goal without any need for a user to directly author computer code. This AI technology permits NLG systems to determine the appropriate content for inclusion in the narrative in a manner that will satisfy the desired communication goal.

Abstract: Techniques are disclosed relating to the serialization of data of a database. A computer system may maintain a database that stores data as a plurality of triples. A triple may include a first data entity that is connected to a second data entity via an edge, and triples of the plurality of triples may be connected to form graphs. The computer system may store metadata defining types of edges between data entities. An edge's type may indicate a procedure for serializing a target data entity connected to that edge's type. The computer system may receive a request to access a data entity associated with a plurality of descendant data entities connected to the data entity via edges. The computer system may generate an output having the data entity and one or more of the plurality of descendant data entities that are serialized according to the types of edges.

Abstract: System and methods are described for anomaly detection and root cause analysis in database systems, such as multi-tenant environments. In one implementation, a method comprises receiving an activity signal representative of resource utilization within a multi-tenant environment; detecting a plurality of anomalies in the activity signal; computing a priority score for each of the plurality of anomalies; correlating at least a subset of the plurality of anomalies to one or more performance metrics of the multi-tenant environment; and transmitting a remediation signal to one or more devices in the multi-tenant environment based on the correlations and the priority scores.

Abstract: The disclosed techniques reduce a responsiveness time for a secondary node state of a database in switching from a second computing node to replace a first computing node acting in a primary node state, with both computing nodes performing the same database queries. The second node receives information regarding queries performed by the first node while in the primary state. In some embodiments, the second node retrieves, from a transaction log, log records detailing operations performed for database transactions. In some embodiments, the second node inserts, based on the log records, data records of the transactions into an in-memory cache of the second node that stores chains of database records from different transactions. Upon receiving sufficient information to switch to the primary state, the second node changes a mode of operation during failover making a committed transaction available for reads by subsequent database queries prior to record reordering.

Abstract: A request to perform an action related to a smart contract recorded on a public trust ledger external to the database system may be received via a communication interface at a database system. The request may authenticate a designated database system account. A determination may be made as to whether the designated database system account is authorized to perform the action by accessing definition information for a public trust ledger synthetic party identifying a plurality of database system accounts authorized to perform one or more actions related to the smart contract. The database system may communicate with a token exchange to perform the action by executing the smart contract when it is determined that the designated database system account is authorized to perform the action.

Abstract: An online system performs replication of database from one system to another. A source database is replicated to a target database by performing an initial bulk copy followed by repeated change data capture. The system performs batching to group events generated by the change data capture events. The system performs a multi-constrained analysis to determine when a batch is ready for sending. The system uses multiple buffers for transmitting events to decouple the receiving and sending of events. The online system may be a multi-tenant system the source and target databases stores data of tenants. The system determines a tenant score to adjust the resources allocated during event transfer. The system splits an event into partial events, to avoid dependency between the partial events so that tasks can be executed independent of each other. Therefore, the system performs replication with reduced synchronization overhead.

Abstract: An AI-enhanced data labeling tool assists a human operator in annotating image data. The tool may use a segmentation model to identify portions to be labeled. Initially, the operator manually annotates portions and once the operator has labeled a sufficient number of portions, a classifier is trained to predict labels for other portions. The predictions generated by the classifier are presented to the operator for approval or modification. The tool may also include an active learning model that recommends portions of the image data for the operator to annotate next. The active learning model may suggest one or more batches of portions based on the extent to which, once labeled, those batches will increase the diversity of the total set of labeled portions.